Daily Edition · Page 6

THE DAILY INDEX

Technology, AI, software, security, crypto and business

MARKETS & BUSINESS

TLDR Crypto

ROBINHOOD TOKENIZES NEARLY 500 US STOCKS, ETFS ON ARBITRUM

Robinhood has deployed 493 tokenized assets worth over $8.5 million on Arbitrum, adding 80 new stock tokens in recent days, including Galaxy, Webull, and Synopsys, with stocks accounting for 70% of tokens and ETFs representing 24%. The tokens are structured as blockchain-based derivatives regulated under the EU's MiFID II rather than representing direct share ownership, offering 24-hour market access, no hidden fees beyond 0.1% FX charges, and a 1 euro minimum investment, though it has drawn scrutiny from Lithuania's banking regulator, who has requested structural clarification.

INTRODUCING PRIVATE MULTISIG

A privacy-preserving multisig wallet implementation combines Merkle tree membership proofs for participant anonymity, aggregated ECC ElGamal encryption for confidential vote casting with bias-free decision making, and Distributed Key Generation eliminating centralized trusted entities, though requiring all participants to vote before revealing outcomes. The system uses babyJubJub key pairs derived from ECDSA signatures via deterministic KDF rather than wallet addresses to reduce identification probability from 1 to 1/N where N is member count, while homomorphic vote aggregation allows summing encrypted votes without decrypting individual ballots until the final decryption key share completes. Limitations include the last voter advantage, where final participants can decrypt prior votes before casting their own, proposal frontrunning possibilities despite deterministic challenge derivation, participant removal deadlocks if removed members don't vote, causing proposals to expire, and linear scalability complexity for vote revelation.

EXPLAINING GKR FOR FAST HASH-PROVING

The GKR protocol turns large “batch and layers” computations like Poseidon2 hashing into simple polynomial evaluations, letting provers commit only to inputs and outputs. With tricks like reducing rounds, batching linear layers, and optimizing sumchecks, GKR achieves around 15x theoretical overhead and under 10x in practice, potentially reaching single-digit overhead. This proving system is the basis behind proposals to reduce the time required to verify Ethereum blocks.

COLLECTIVE INTELLIGENCE AND PREDICTION MARKETS

Prediction markets transform collective intelligence into a structured, incentivized system where people are rewarded for accurate insights. Unlike open-source projects or Wikipedia, which rely on voluntary contributions, prediction markets aggregate diverse information through financial stakes, encouraging better, data-driven forecasts. By giving experts and informed outsiders monetary incentives to share knowledge, these markets create more accurate crowd predictions. There are remaining challenges around liquidity and market diversity, but crypto-based prediction platforms are closing the gap between human collaboration and actionable forecasting.

TLDR IndexPage 6